DB / Oracle / 备份与迁移

Oracle Data Guard 架构详解

2026.09.06

Oracle Data Guard 架构详解

适用版本：Oracle Database 10g / 11g / 12c / 19c / 23ai 文档版本：v1.0 / 2026-07

1. 概述

Data Guard 是 Oracle 灾备方案[1]：

类型：

  • 物理 Standby
  • 逻辑 Standby
  • Snapshot Standby
  • Far Sync

详细见：Oracle Data Guard 部署。

2. 架构

2.1 主库

- Primary
- 读写
- Redo 产生

2.2 备库

- Standby
- 物理：块级一致
- 逻辑：SQL 应用

2.3 角色

- Primary：主
- Standby：备
- 可切换

3. 数据保护模式

3.1 Maximum Protection

- 零数据丢失
- SYNC
- 备库失败主库停

3.2 Maximum Availability

- 零数据丢失（最佳）
- SYNC
- 备库失败主库继续

3.3 Maximum Performance

- 最大性能
- ASYNC
- 数据可能丢失

3.4 设置

ALTER DATABASE SET STANDBY DATABASE TO MAXIMIZE PROTECTION;
ALTER DATABASE SET STANDBY DATABASE TO MAXIMIZE AVAILABILITY;
ALTER DATABASE SET STANDBY DATABASE TO MAXIMIZE PERFORMANCE;

4. Redo 传输

4.1 SYNC

- 同步
- 主库等备库确认
- 零丢失

4.2 ASYNC

- 异步
- 主库不等
- 性能好

4.3 配置

ALTER SYSTEM SET log_archive_dest_2 = 
  'service=standby sync valid_for=(online_logfiles,primary_role) db_unique_name=standby'
  SCOPE=BOTH;

5. 物理 Standby

5.1 创建

-- 1. 主库准备
ALTER DATABASE ARCHIVELOG;
ALTER DATABASE FORCE LOGGING;
ALTER DATABASE ADD STANDBY LOGFILE GROUP 11 ('...') SIZE 200M;

-- 2. 参数
ALTER SYSTEM SET log_archive_config = 'dg_config=(orcl,orcldg)';
ALTER SYSTEM SET log_archive_dest_2 = 'service=orcldg async ...';

-- 3. RMAN Duplicate
rman target sys/pwd@orcl auxiliary sys/pwd@orcldg
DUPLICATE TARGET DATABASE FOR STANDBY FROM ACTIVE DATABASE DORECOVER NOFILENAMECHECK;

-- 4. 启动 MRP
ALTER DATABASE RECOVER MANAGED STANDBY DATABASE DISCONNECT FROM SESSION USING CURRENT LOGFILE;

5.2 应用

- Redo Apply
- MRP（Managed Recovery Process）
- 物理块级

5.3 Active Data Guard

- 11g+
- 只读 + 应用
- 实时查询

详细见：Oracle Active Data Guard。

6. 逻辑 Standby

6.1 创建

-- 类似物理
-- 但使用 SQL Apply
ALTER DATABASE START LOGICAL STANDBY APPLY;

6.2 应用

- SQL Apply
- LogMiner
- 转换 SQL
- 执行

6.3 限制

- 数据类型限制
- 不支持对象
- 需要主键

7. Snapshot Standby

7.1 转换

ALTER DATABASE RECOVER MANAGED STANDBY DATABASE CANCEL;
ALTER DATABASE CONVERT TO SNAPSHOT STANDBY;

7.2 用途

- 读写测试
- 临时
- 恢复后同步

7.3 恢复

ALTER DATABASE CONVERT TO PHYSICAL STANDBY;
ALTER DATABASE RECOVER MANAGED STANDBY DATABASE DISCONNECT;

8. Far Sync

8.1 概述

- 12c+
- 远距离零丢失
- 中继实例

8.2 架构

Primary → Far Sync (SYNC) → Standby (ASYNC)

8.3 优势

- 零数据丢失
- 远距离
- 性能好

9. 切换

9.1 Switchover

- 计划
- 主备互换
- 无数据丢失
-- 主库
ALTER DATABASE COMMIT TO SWITCHOVER TO STANDBY WITH SESSION SHUTDOWN;

-- 备库
ALTER DATABASE COMMIT TO SWITCHOVER TO PRIMARY;
ALTER DATABASE OPEN;

9.2 Failover

- 故障
- 备库变主
- 可能数据丢失
-- 备库
ALTER DATABASE RECOVER MANAGED STANDBY DATABASE FINISH;
ALTER DATABASE ACTIVATE STANDBY DATABASE;

9.3 Reinstate

- 12c+
- 失败主库恢复为备
DGMGRL> reinstate database orcl;

10. Data Guard Broker

10.1 配置

# 启用
ALTER SYSTEM SET dg_broker_start = TRUE;

# DGMGRL
dgmgrl sys/Oracle123@orcl
CREATE CONFIGURATION dg_config AS PRIMARY DATABASE IS orcl CONNECT IDENTIFIER IS orcl;
ADD DATABASE orcldg AS CONNECT IDENTIFIER IS orcldg MAINTAINED AS PHYSICAL;
ENABLE CONFIGURATION;

10.2 管理

SHOW CONFIGURATION;
SHOW DATABASE orcl;
SHOW DATABASE orcldg;

EDIT DATABASE orcldg SET PROPERTY LogXptMode = 'SYNC';
EDIT CONFIGURATION SET PROTECTION MODE AS MAXAVAILABILITY;

SWITCHOVER TO orcldg;
FAILOVER TO orcldg;

11. Standby Redo Log

11.1 创建

ALTER DATABASE ADD STANDBY LOGFILE GROUP 11 
  ('/u01/oradata/stdby_redo11.log') SIZE 200M;
ALTER DATABASE ADD STANDBY LOGFILE GROUP 12 
  ('/u01/oradata/stdby_redo12.log') SIZE 200M;
ALTER DATABASE ADD STANDBY LOGFILE GROUP 13 
  ('/u01/oradata/stdby_redo13.log') SIZE 200M;
ALTER DATABASE ADD STANDBY LOGFILE GROUP 14 
  ('/u01/oradata/stdby_redo14.log') SIZE 200M;

11.2 优势

- 实时应用
- 零丢失
- Active Data Guard

详细见：Oracle Redo 日志配置与管理。

12. 监控

12.1 v$archive_dest

SELECT dest_id, status, destination, error 
FROM v$archive_dest WHERE dest_id <= 5;

12.2 v$archive_dest_status

SELECT 
  dest_id, database_mode, 
  recovery_mode, 
  archived_seq#, applied_seq#
FROM v$archive_dest_status;

12.3 v$dataguard_status

SELECT facility, severity, message, timestamp
FROM v$dataguard_status
ORDER BY timestamp DESC;

12.4 进度

-- 主库
SELECT thread#, sequence#, archived, status FROM v$log;

-- 备库
SELECT thread#, sequence#, applied, status 
FROM v$archived_log ORDER BY sequence# DESC;

12.5 Gap

SELECT * FROM v$archive_gap;

13. 性能优化

13.1 Redo 传输

- 网络带宽
- ASYNC
- Net Timeout

13.2 应用

- Parallel
- SGA
- 备库性能

详细见：Oracle Data Guard 性能优化。

14. 常见坑与排错

14.1 Gap

- 归档缺失
- FIND_GAP
- 手动传输
- REGISTER

14.2 ORA-16191

- SYS 密码不一致
- 密码文件同步

14.3 ORA-12154

- 网络问题
- tnsnames
- 监听

15. 最佳实践

  1. 物理 Standby：简单
  2. Active Data Guard：实时查询
  3. Maximum Availability：平衡
  4. SYNC：关键业务
  5. Standby Redo Log：实时
  6. Broker：管理
  7. Far Sync：远距离
  8. 监控 Gap：及时
  9. 测试切换：可行
  10. 文档化：DR 计划

16. 参考资料

[1] Oracle Data Guard Concepts and Administration 19c https://docs.oracle.com/en/database/oracle/oracle-database/19/dgbkr/